Oct 2019 No.439

Safety at Sea during the Industrial Revolution

Morgan Kelly, Cormac Ó Gráda and Peter Solar

WORKING PAPER SERIES

Centre for Competitive Advantage in the Global Economy

Department of Economics

Safety at Sea during the Industrial Revolution

Morgan Kelly, University College Dublin Cormac Ó Gráda, University College Peter Solar, Vesalius College, VUB, Brussels

ABSTRACT: Shipping was central to the rise of the Atlantic economies, but an extremely hazardous activity: in the 1780s, roughly five per cent of British ships sailing in summer for the United States never returned. Against the widespread belief that shipping technology was stagnant before iron steamships, in this paper we demonstrate that between the 1780s and 1820s, a safety revolution occurred that saw shipping losses and insurance rates on oceanic routes almost halved thanks to steady improvements in shipbuilding and navigation. Iron reinforcing led to stronger vessels while navigation improved, not through chronometers which remained too expensive and unreliable for general use, but through radically improved charts, accessible manuals of basic navigational techniques, and improved shore-based navigational aids.

Keywords: shipping, insurance, Industrial Revolution JEL Classifications: N, N73, G22

1 "Curse thee, thou quadrant!" dashing it to the deck, "no longer will I guide my earthly way by thee; the level ship’s compass, and the level dead-reckoning, by log and by line; these shall conduct me, and show me my place on the sea.

Captain Ahab in Moby Dick, Ch. CXIII

It is hard to imagine any narrative of the development of western Europe—what Ralph Davis (1972) termed "The Rise of the Atlantic Economies"—where shipping is not central, from the Voyages of Discovery, through the trade in tobacco, sugar, and slaves, to the fact that every ounce of cotton spun in Manchester was imported, and much of its end product subsequently exported. In the 1780s Britain had roughly 50,000 merchant seamen and one million tons of shipping, and by 1831 these had risen to 130,000 men and 2.5 million tons.1 However, alongside its political and economic importance, sailing was a hazardous activity. During the 1780s, for example, about five per cent of British ships leaving in summer for the United States never returned, and for longer voyages casualties were even higher. Going back to the work of North (1968) and Harley (1988), the consensus among economic historians has been that shipping technology was largely stagnant before the appearance of iron steamships in the mid- nineteenth century. In this paper we show instead that between the 1780s and 1820s a revolution in safety took place at sea that saw a halving of

1 Earle 1998, p. 7; United Kingdom 1831, p. 14; Phillips 1832, p. 676-8.

2 losses on oceanic routes, with smaller safety improvements on European routes. The paper falls into two parts. First we calculate the loss rates of shipping during peacetime periods on various routes at different times between the 1730s and 1830s, both directly from recorded casualties, and indirectly from the premiums charged to insure voyages. In both cases we find sharp falls in the decades around 1800. We then seek to explain the sources of these radical improvements by laying out the numerous improvements in shipbuilding and navigation that occurred at this time. We collect peacetime insurance rates for summer sailings for the 1730s, 1760s, 1780s, 1820s and 1840s for oceanic routes (North America, Africa and the Caribbean, and Asia) and European ones (the Baltic, North Sea, Bay of Biscay, and Mediterranean). For ship losses, we construct loss rates on each route for the 1780s and 1820s. Our insurance data are taken from a variety of sources—including newspaper accounts, brokers’ risk books, and institutional histories—while our direct loss data come from Lloyd’s List. What we discover is that the insurance and loss series are, in most cases, both internally consistent, and consistent with each other. Insurance rates on different routes change together consistently, as do loss rates. Most importantly, on each sailing route, with a few easily explained exceptions, changes in insurance rates and loss rates track each other closely. Although any single insurance or loss figure taken in isolation might potentially be problematic, the fact that all our data move closely together suggests that they are broadly reliable, and capture a real safety revolution that took place in the decades around 1800. Losses and premiums on oceanic routes roughly halve between 1780 and 1830, with lower falls on most shorter routes within Europe. Driving this safety revolution were a broad range of gradual improvements both in the seaworthiness of ships and in their navigation. Looking at seaworthiness, the central advance stemmed from the growing availability of cheap, high quality metals. The diffusion of copper sheathing from the 1780s preserved the structural integrity of hulls by protecting them against shipworm. The increased supply of wrought iron allowed the flimsy wooden brackets used to hold ships together to be increasingly replaced by iron fastenings, making for sturdier ships that were less prone to leak or disintegrate in heavy seas.

3 As for navigation, the popular perception (shared even by Landes 1983, 145-157) that the decisive breakthrough came with the determination of longitude with chronometers is simply wrong. Chronometers remained too expensive and unreliable to be adopted on ordinary merchant ships before the 1840s. The important changes instead were greatly improved charts, the growing availability of affordable manuals teaching basic navigational techniques, and improved navigational aids such as lighthouses and channel markings. In general, what we find is that the larger ships on long distance routes experienced the greatest improvements in safety. The paradigmatic example of technological innovation is the East India Company which had the incentives and financial resources to pioneer improvements in shipbuilding and navigation on its large and expensive ships. In the 1780s, despite the far longer and more hazardous voyages it undertook, its loss rates were no higher than on the route to the United States, and by the 1820s its losses were similar to those on vessels going to the Mediterranean. In terms of the existing literature, the evolution of British insurance rates during the Industrial Revolution has received little attention outside the work of Leonard (2012, 2013, 2014, 2016) and the recent study of the slave trade by Pearson and Richardson (2019). The revolution in maritime safety between 1780 and 1830 appears to have gone unnoticed. The rest of the paper is as follows. First, we analyse the changes in insurance rates that occurred during the Industrial Revolution. Then we take data from Lloyd’s List to show how their evolution tracks falls in loss rates on corresponding shipping routes. The following sections describe the advances in shipbuilding and navigation that underlay this safety revolution. The final section ties our findings of rapid technological advance in ocean shipping to the wider literature on the Industrial Revolution.

Marine Insurance Rates The routine hazards of shipping prompted risk sharing mechanisms that by the mid-fourteenth century had evolved into more or less their modern form of brokers providing cover for a premium set at a percentage of the amount insured.2

2 The classic reference on the early evolution of insurance is Edler de Roover (1945). See also Kingston (2013) and Leonard (2014).

4 In London a sophisticated marine insurance industry developed during the seventeenth and eighteenth centuries. From 1720 two large chartered companies, Royal Exchange Assurance and London Assurance, which between them accounted for about 10 per cent of business, coexisted with the hundreds of individual underwriters working at Lloyd’s and other coffee-houses, the coal exchange, brokers' offices, their own counting houses, and elsewhere. In the eighteenth century two publishing enterprises grew up to serve this industry. Lloyd’s Register met the need for reliable information on ship characteristics, masters, and owners by producing annual lists of vessels susceptible to be insured; and Lloyd’s List, a newspaper published twice weekly, disseminated information on ship movements and incidents at sea. Besides the destination of the voyage, the time of year, the deductible charged, and whether peace or war prevailed, insurance rates reflected many other considerations such as the type of vessel, its cargo, and the reputation of the owner and master. It is therefore not surprising that surviving rates for individual voyages to the same destination can vary substantially. A notable feature of these markets is how rapidly premiums could change in response to new information about risks. In 1808, in response to the information that a French squadron under Villeneuve had arrived in the West Indies, homeward rates from there jumped from 8.5 to 15 per cent. These rose to 16 per cent with news that he was heading for the English Channel, but fell back to 11 per cent after his indecisive encounter with Calder and return to Cadiz (Wright and Fayle, 1928, p. 188). Our analysis of insurance rates for peacetime periods from the 1730s to the 1820s is based on a variety of sources.3 For 1730-1731 and 1768-70 our rates are taken from John (1951, Table 2), using the outward rate charged by the London Assurance Company. For the case of the East Indies we use half the round trip rate posted. In 1763–70 the rates are based on 1,368 individual contracts from the risk book of merchant-insurer William

3 Wartime rates could be much higher: in the late 1790s the rates to Jamaica were three times higher in convoy and six times higher without convoy; the rates to the eastern Mediterranean, when not quoted as “uncertain”, were more than ten times higher (Prince’s Price Current, 1796- 1799). Naturally, “peacetime” in the eighteenth century was a fairly elastic concept, understood by contemporaries as a temporary break from the usual state of continuous warfare. Contracts took this into account: for the lengthy round trip to the East Indies around 1770, the London Assurance Company charged £15 “with £8 to be refunded if no hostilities in European waters” (John 1951, Table 1).

5 Braund.4 In later years some of our rates are based on averages of mid- month quotations in price currents and newspapers. For 1783–1785 we rely on Prince’s Price Current; and for 1820-1829 on London New Price Current. The premiums are the median quotation for voyages from London between May and September, the period during which most journeys took place. Finally, for 1828–1830 we have the summer rates calculated for Lloyd’s of London by Wright and Fayle (1928, 319–320).5

[Table 1 about here]

These insurance rates are shown in Table 1. It can be seen immediately that the rates up to the 1780s on different routes are, with a handful of exceptions, close to each other, as are those for the 1820s, with a notable fall occurring in between. The most notable exceptions are for the Mediterranean but, as Figure 2 shows, this was a relatively thin market. The consistency of rates across the larger routes suggests that the numbers are broadly reliable, something that we confirm further below by comparing them with actual loss rates on these routes. The second thing that is clear is that premiums are unchanged between the 1730s and 1780s, but then fall by the 1820s on most routes, especially on oceanic ones. Some things should be noted. The first is that although the Baltic is, alongside the North Sea, the most important route (see Figure 2 below), in many cases the premium is not reported, presumably because it was a fixed markup on the Hamburg rate. Second, the stability of the East Asia rates is misleading. In the eighteenth century they refer to the ships of the East India Company, but by the 1820s the vast majority of ships sailing to India were smaller, less sturdy vessels belonging to individual shipowners.

4 Essex County Record Office, D/DRu B7, William Braund’s Journal of Risks, 1759-1765. We are grateful to Adrian Leonard for sharing these data with us. 5 After the late 1820s insurance premia were rarely quoted in newspapers or price currents. For 1842-1844 we have a few rates quoted irregularly in the Public Ledger and Daily Advertiser, but in many cases the range of rates quoted to particular destinations is quite large and larger than the ranges in the 1820s, with the result that average rates appear to be higher. Specifically, the percentage rates are Hamburg 1.25; Biscay 0.75; Lisbon 1.12; Leghorn 1.75; Smyrna 2.0; Canada 2.25; America 1.50; Jamaica 1.37; East Indies 2.25. MacGregor (1847, 181) reports minimum summer rates from New York to Europe in 1840 of 0.75 per cent, with rates in the opposite direction running from 0.75 to 1.75 per cent.

6 [Figure 1 about here]

The evolution of rates is shown in Figure 1 which graphs rates quoted for the London Assurance Company in the 1730s and 1760s and Lloyds in the 1780s and 1820s. The figure highlights the stability of rates through before the Revolutionary and Napoleonic wars, and the steep fall afterwards. Rates to America halve, and those on European routes fall by over a quarter. We may further gauge the reliability of our insurance data by comparing them with rates quoted in other European markets. From 1766 to 1780 we have monthly data for Amsterdam compiled by Spooner (1983), and these are close to London rates to the same destinations: 1 to 1.5 per cent on northern routes, and 2 per cent on Mediterranean ones, with rates of around 2.5 to 4 per cent on the only oceanic routes listed, to Curacao and Surinam. Rates are also available for northern Europe and the Mediterranean from Hamburg from the 1730s to the 1850s (Denzel 2017).6 As for the London rates, the Hamburg rates remain fairly constant, outside wartime, during the eighteenth century, though they are notably higher than London rates, possibly reflecting a thinner market with a shallower capital base and fewer opportunities for risk sharing.7 The rate from Hamburg to London hovered around 1.5 per cent, compared with the London rate of 1 per cent. The average peacetime rate across various destinations varied between 2.5 and 3 per cent. After the wars Hamburg rates were lower and trending downwards. The London rate had fallen to one per cent by the early 1820s and dipped below that by the early 1830s. The average rate fell to about 2 per cent in the 1820s and reached 1.5 per cent by the early 1840s. The reductions in rates on European routes from the 1780s to the 1820s in the Hamburg insurance market are thus comparable to what we find for the London market.

6 These trends in maritime insurance rates differ from those described by Chet (2014), but his summary of rates comes from a variety of disparate and perhaps incomparable sources. 7 The unique depth of its insurance market that allowed overseas trade to continue during wartime was another of eighteenth century Britain’s “sinews of power”, whose strategic importance was recognized in repeated government efforts to prevent Lloyd’s from insuring enemy shipping (Leonard 2013; Clark 2004).

7 Institutional Changes Marine insurance rates may have fallen for several reasons other than improving maritime safety.8 One possibility would be innovations in underwriting. Leonard (2014, ch. 5) argues, however, that by the 1780s the institutions of the modern British marine insurance industry were already in place. Underwriters had left the coffee houses for purpose-built premises and had established rules for self-regulation of their activities. The Common Law had been adapted, in a process led by Lord Chief Justice Mansfield and relying on evidence of merchant and underwriter practice, to provide an effective vehicle for resolving disputes with outsiders. Leonard sees very little change in this institutional environment over the French wars, although high wartime premiums did lure much new capital, some of it rather “naïve”, into the business. In 1824 Parliament did enact legislation authorising the creation of additional joint stock insurance companies, but new entrants were too late to have influenced the earlier fall in marine insurance rates documented above. Moreover, the entry of new firms does not seem to have had any immediate effects since insurance premiums in the late 1820s were not significantly lower than in the early 1820s. Even without significant institutional change, it is possible that marine insurance rates fell because the London insurance market became more competitive between the 1780s and the 1820s. When New Lloyd’s was launched in 1771, there were 71 subscribers, though many underwriters remained in other premises throughout the City. By 1810 the number of subscribers had risen to 1,400 or 1,500, of whom perhaps two-thirds were underwriters (United Kingdom 1810, p. 107) and to over 1,850 in 1815 (Wright and Fayle, 1928, 320). One broker with long experience of the market testified to the 1810 committee inquiring into marine insurance that “competition with the common underwriters has lowered premiums so much, that the best underwriters have been obliged to discontinue the business they lower the premiums upon one another so much” (United Kingdom 1810, p. 110).

8 For earlier periods, Leonard (2016) attributes the large falls in London rates on European routes from about 8 per cent around 1600 (a figure that Puttevils and Deloof (2017) also find for Antwerp) to under 2 per cent by 1760 to a larger, more competitive market with increasing liquidity allowing greater diversification of risk, where brokers were better able to calculate loss probabilities.

8 One source of intensified competition was from overseas exchanges, notably Hamburg. British insurance contracts faced extremely heavy stamp duty with European routes subject to a 2s 6d duty on the amount insured, and longer ones a 5s duty; adding an extra 20 per cent to the cost of insuring a voyage to the Baltic or United States. In Hamburg, by contrast, stamp duties were one sixth of British levels. Combined with a severe postwar depression in shipping this caused the membership of Lloyd’s to fall by one third by 1830, and stamp duty receipts to almost halve compared with the last wartime years. (Wright and Fayle, 1928 320).

We can see in Figure 2 that the rates for Hamburg and the Baltic in the 1820s diverge markedly from those for Lisbon, and fall more than the changes in loss rates (see Figure 3) would lead us to expect. We suspect that increased competition accounts for some of this change. However, when it comes to oceanic routes, these are barely mentioned in Denzel’s (2017) account of the Hamburg market, suggesting that competition was less of a factor there, and we will see below that premiums to the United States change by the same proportion as loss rates.

Loss Rates Marine insurance could be taken out on the value of the ship or on the value of the cargo or on both. Our source for losses, Lloyd’s List, the current incarnation of which began publication in 1734, records many incidents in which cargo was damaged, waterlogged, washed away or jettisoned to save the ship. However, since we suspect that Lloyd’s List was less complete on cargo losses than on ship losses and we have no information on either the values of the cargo lost or the total value of cargo insured, we focus on ship losses. For ships we lack information on the tonnage of the vessel or its value, which could range from a few hundred pounds for small, old vessels to tens of thousand pounds for East Indiamen. Instead we count the numbers of ships losses recorded in the marine news columns of Lloyd’s List.9

9 On the early history of Lloyd’s List see McCusker 1991. Records of losses were also kept directly by Lloyd’s since 1774; unfortunately, the early Loss Books were destroyed by a fire in the Royal Exchange in 1838. The U.K. government only started publishing information on ship wrecks from the 1840s. One series, with only aggregate figures, drew upon losses recorded in the official ship registers; the other, from the early 1850s with much more detail, collated information from Coast Guard officers, Admiralty legal records, returns made to insurers at Lloyd's, and reports in Lloyd's Lists and the Shipping and Mercantile Gazette.

9 It might appear that the official ship registers kept in the ports of Britain and Ireland since 1786 could offer additional or even better information on losses. These registers recorded the owners, masters and dimensions of ships, and in principle they recorded when and how a ship went out of registration, whether it was transferred to another port, sold abroad, broken up or lost. However, besides the practical problems of using these numerous, physically massive registers, they are frequently uninformative about ship losses. Often no date is given, and only rarely do they provide detail on where and how the ship was lost, or in which trade it was being used at the time. In fact, as will be seen below, many ship losses in the foreign trade of Britain and Ireland are simply not reported in official registers. Counting ship losses in Lloyd’s List is not without its own problems. A typical entry in its Marine List section is a snippet that includes the ship’s name, its master’s name, its ports of origin and destination, and the location and nature of the incident, though sometimes the port to which it belonged was given instead of or in addition to those for origin and destination. Yet sometimes much of this information is missing, particularly when the report was based on another ship’s observation of wreckage. Ships were variously described as lost, abandoned, missing, wrecked, burnt and on shore. Not all of these incidents turned out to be definitive losses. Ships that were missing later turn up in port, and some ships that were abandoned were subsequently towed in. The most problematic reports concern ships described as on shore. Many are later reported to have been got off, sometimes pulled off by other vessels or boats, sometimes floated off at higher tides, aided in some cases by offloading cargo to lighten the ship. But it is clear that Lloyd’s List did not report all of these rescues. We have successfully matched over half of the ships recorded as lost to entries in the annual lists of ships in Lloyd’s Register in the year before the ship was lost. We then searched for the same ships in Lloyd’s Register two years later, with the result that about a quarter of the matched ships turned out to be still in service, including a few initially reported as wrecked. These survivors were removed from the data set. How complete was Lloyd’s List as a record of ship losses? Data collected for the 1836 Parliamentary Select Committee on Shipwrecks provide a way to approach this question. The report’s appendices contain two lists: one is described as a "return of all vessels belonging to the United Kingdom, reported on the books of Lloyd's as missing or not heard of, during the years 1833, 1834 and 1835"; the other as a "return of all ships' registers cancelled or given up on account of the loss or destruction of the ships to which they belonged", also for 1833 to 1835. The first turns out to have been a quite accurate extraction of data from Lloyd's Lists,

10 with careful attention having been paid to excluding both foreign vessels and vessels that were not conclusively lost. The second list comes from the official ship registers mentioned above. These two contemporary lists of ship losses can be compared with a modern compilation to be found on Wikipedia. The Wikipedia lists of wrecks are meticulously referenced, cover a long period and refer, at least in early years, mainly to British vessels. Until the mid-1820s they are based primarily on Lloyd's Lists, and hence do not differ significantly from our principal source; thereafter they draw on both Lloyd’s List and a wide range of other newspapers, picking up losses that may not appear in Lloyd's Lists. The Wikipedia 1834 list thus constitutes an alternative, albeit not entirely independent, record of ship losses. All three sources arrive at a similar numbers of losses (Table 2), but they turn out not to be all the same losses. Across the three sources there were over a thousand distinct losses, which was 77 to 103 per cent more than the number in any individual source. The major discrepancy concerns losses that showed up only in the official ship registers, and these were predominantly small vessels involved in the coasting trade. To a lesser extent, coasting vessels also accounted for many of the losses reported only in the Wikipedia compilation or in both Wikipedia and the registers. By contrast, the losses reported in Lloyd's Lists consistently tend to be larger vessels involved in foreign trade, though coasting vessels were still fairly prominent in this source.

[Table 2 about here]

A striking result of this exercise is the large number of losses reported in Lloyd's Lists and Wikipedia that did not show up in the official ship registers. It is possible that some of these observations may refer to foreign rather than British vessels, and hence would not have been in the official ship registers. Others may have been vessels that were refloated but not reported as such in the press. But the number is so large that it would seem that the official registers must have missed many losses, in particular those of larger vessels on foreign routes. In such cases news of the ship's fate may not have come back to the local port authorities, or may not have come back in a timely or accurate way. Analysis of the official ship registers at the port of Whitby in Yorkshire from the late eighteenth to the mid- nineteenth century shows that very often ships were recorded as lost, but the date at which the loss took place was not specified. Sometimes ships that had been lost remained on the official registers until there was a periodic clean up.

11 We focus on loss rates in the foreign trade of Britain and Ireland, omitting the coastal trade, for two reasons. First, as just shown, Lloyd’s List’s coverage of losses in the coasting trade is very incomplete. That of the foreign trade is also incomplete, but not to the same extent. In 1834 420 vessels that were lost could be identified as in the foreign trade, of which Lloyd’s List recorded two thirds. This share may be understated because the Wikipedia list may be less than assiduous in picking up vessels that were refloated, and for want of information some vessels may be double-counted because they have not been successfully matched across sources. Nonetheless, because Lloyd’s List certainly missed some losses in the foreign trade our loss rates will be underestimates. Second, in order to calculate loss rates we need a measure of the number of voyages susceptible to loss. From the late eighteenth century there are statistics of the number of ships entering and clearing British and Irish ports, but only from the early nineteenth century do these statistics include voyages in the coasting trade. Throughout the period the entries and clearances are given by foreign country, so they can be matched reasonably well to the destinations for which marine insurance rates were quoted. One can only speculate on how the extent to which Lloyd’s List coverage of losses in the foreign trade may have changed over time. If it fell from the 1780s to the 1820s, then any fall in loss rates would overstate the true fall. But there are good reasons to think that any bias would be in the other direction, toward understating any true fall. The thoroughness with which Lloyd's List recorded ship losses is likely to have increased over time as its network of correspondents became denser. In 1785 departures and arrivals were reported for about 50 different British and Irish ports and about a hundred foreign ports; by 1825 both numbers had roughly doubled.10 In addition, the number of column inches devoted to marine incidents more than doubled.

[Table 3 about here]

We have extracted ship losses from Lloyd’s Lists and calculated loss rates for peacetime periods in the mid-1780s and the mid-1820s (Table 3). Since the insurance rates refer only to the outward voyage, the losses were divided by two because they refer to both outward and return voyages. The divisor is the larger of the numbers of entries or clearances for each foreign region. On some routes these diverged significantly because ships either arrived or departed in ballast and were

10 Based on the numbers of distinct ports reported in January and July of both years.

12 not counted in the statistics. The extreme was the Baltic, for which entries exceeded clearances by over threefold in the 1780s and by 75 in the 1820s. Even in ballast, ships would still need to be insured, so they should be counted in the denominator. The resulting loss rates are plausible in that, like the marine insurance rates, they increase with the length of the voyage. This approach to counting losses is uninformative on the Asia route: based on fewer than 100 arrivals recorded in 1783-85 we cannot be confident of the true loss rate. We therefore prefer losses per voyage over longer intervals calculated directly from the records of East India Company ships: these fell from 1.7 per cent in 1765-73 and 1.3 per cent in 1783-91 to 0.6 per cent in 1820-9 (Solar 2013). In other words, the loss rates were smaller than those on the much shorter and more straightforward route to the United States, which did not entail rounding the Cape of Good Hope and, for ships going to China, passing through the various straits in the Indonesian archipelago. These relatively low losses suggest that the East India Company was significantly ahead of ordinary merchant shipping in adopting advances in shipbuilding and navigation: in fact, as we will see below, several of the most important advances were actually pioneered by the Company.

[Figure 3 about here]

The loss rates match up reasonably well to the insurance rates. In most cases the loss rate is less than the insurance, as one would expect if the insurers were to make a profit. However, given the sources, this did not necessarily have to be the case. Our loss rates are not always a perfect match to the insurance rates. For example, the rates quoted to the West Indies almost certainly referred to the direct voyages, but our loss rates include the much riskier voyages that stopped in Africa to collect slaves. The rates that insurers actually quoted customers on a given route depended on the characteristics of particular ships and routes and could vary considerably, with a long tail corresponding to poor risks. The range of rates quoted by the Royal Assurance, Lloyd’s and in the newspapers probably referred to better risks, whilst the calculated loss rates include ships of all sorts and could thus be higher. Also, as in any contract prone to substantial moral hazard, marine insurance policies typically involved deductibles (“abatements”), which meant that the insurers paid out less than the full value insured: for the Royal Assurance the deductible was 16 per cent (John, 1951). Finally, as insurers could earn income by investing premiums, there was some scope to reduce insurance rates somewhat. Loss rates to most destinations fell significantly from the 1780s to the 1820s, and the extent to which they fell matches reasonably well the fall in marine

13 insurance rates. This suggests that falls in insurance rates were largely driven by improved safety, rather than intensified competition. The two glaring exceptions are the Baltic and Canada, where the loss rates rose by half, but each of these exceptions can be explained by changes in the nature of the trades. The Mediterranean and East Indies trades present lesser, though also explicable, anomalies. The rise in the Baltic loss rate reflects a change in the seasonality of the trade. The great danger in this trade was winter ice, to such an extent that when insurance rates for December, January and February were cited (and they often were not), they were sometimes ten times the summer rate; by November they were already three times as high. What seems to have happened between the 1780s and the 1820s is that shippers stretched the sailing year. This can be seen in the dates at which losses in the Baltic were reported in Lloyd’s List. In the mid- 1780s 33 per cent of losses were reported between December and April; in the mid-1820s this has risen to 62 per cent. It can also be seen in arrivals of ships in London from the Baltic: in 1789 only 3 per cent arrived in the six months from December to May; in 1826 this had risen to over 20 per cent (Lloyd’s List 1789, 1826). Since losses were still relatively higher in the winter months, this shift in seasonality would tend to produce increased loss rates. The rise in the Canadian loss rate reflects the growth of the dangerous timber trade which appeared when supplies of Baltic timber were first cut off during the Napoleonic wars, and afterwards faced increased tariffs. The losses on the Canadian route—where old vessels carrying a low-density cargo, some of it loaded on deck, made them prone to capsize—drove one of the earliest pieces of industrial safety legislation, that banning deck cargoes in 1839 (Williams 2000). According to the parliamentary inquiry that preceded this law, in the period 1836–1838, 226 ships were lost heading from Canada, a rate of 4.2 per cent (this is an average for all seasons of the year), and crews in vessels that became waterlogged but remained afloat suffered severe privations sometimes leading to cannibalism (United Kingdom 1839, pp. i-v). Of the roughly 2,000 ships coming from Canada each year, 87 per cent were carrying timber. Crews on these vessels earned 10s per week compared with 7s on other ships, and insurance rates ran from 3-3.5 per cent in spring to 10 per cent in autumn. By contrast reported premiums for the Baltic trade varied from 0.5 per cent in summer to 8 per cent in winter. This suggests that the insurance rate of 1.8 per cent for the 1820s in our Table 1 reflects a mixture of high-risk timber ships alongside ordinary vessels paying similar rates to those for America.

14 The loss rate on the Mediterranean trade fell by much more than did insurance rates to Leghorn and Smyrna; however, insurance rates in the mid-1820s may have been kept up by the fear of losses to shipping caused by the struggle for Greek independence and by the presence of South American privateers off the Spanish coast. More likely however, is that the discrepancy reflects the fact that the Mediterranean was a relatively minor route plied by small vessels. As noted above, loss rates on East India Company ships fell from the 1780s to the 1820s along with other rates. However, the marine insurance rates shown in Table 1 remained stable, a discrepancy that we believe arises from a change in the composition of ships in the India trade. In the 1780s all ships were operated by the East India Company; by the 1820s the vast majority of ships were not Company vessels. The end of the India monopoly in 1814 opened the trade to smaller, less heavily armed and manned vessels, which had higher loss rates than did Company ships. The rate calculated from Lloyd’s List for ships bound to Asia in the mid- 1820s is 3.1 per cent; the loss rate on Company ships was 0.6 per cent in 1820-9. One notable feature of the evidence from Lloyd’s Lists is the rarity of losses of British ships to pirates both in the 1780s and the 1820s. The fact is that large scale piracy had been disappeared considerably earlier: at their peak in the 1620s and 1630s, the Barbary corsairs were taking only around 15 English ships a year, and subsequent treaties protected British ships from being taken. The later upsurge in Caribbean piracy after 1716 was eradicated within a decade with 400-600 pirates hanged (Earle 1998, 119–120; Rediker 1987, 256). Wartime privateering was a different matter. Wright and Fayle (1928, 187-190) estimate that the risk of capture from 1793-1801 averaged about 3.4 per cent per voyage but varied substantially by year and route. They reproduce one broker’s rates for 1808 where some summer premiums to the Americas go as high as 12 per cent, and to the West Indies up to 20 per cent. For the Baltic and “Dutch Ports in Enemy’s Hands” premiums ran up to 40 per cent.11 But our concern here has been with peacetime rates, which are more germane to longer-term changes in safety.

11 Chet (2014) has argued that piracy persisted into the nineteenth century, and the Lists do indeed report ships of other nations being taken by pirates or privateers. In the mid-1780s the “Algerines” are recorded as capturing American, Portuguese, Venetian, Neapolitan and Genoese ships, but no British ships, holders of Mediterranean passes, were taken. In the mid-1820s the independence struggles of the Greeks and South Americans rendered dangerous the eastern Mediterranean and the Atlantic as far as the Spanish coast; however, British ships do not appear to have been interfered with. Hence, it is unlikely that the elimination of piracy had much to do with the fall in peacetime insurance rates.

15 Technological Innovation in Shipbuilding A widespread view is that by the late eighteenth century the European sailing ship was a mature technology and that most subsequent innovations were incremental. In fact, several structural innovations occurred in shipbuilding that served to reduce losses, notably copper sheathing, iron reinforcing, and flush decks. Copper sheathing was introduced in the 1780s and spread rapidly from the slave trade to the East India and West Indies trades (Solar 2013; Solar and Rönnbäck 2014). By protecting hulls from the depredations of ship worm, coppering prolonged ship lives and reduced maintenance (Solar 2013; Solar and Rönnbäck 2014). A sound hull, moreover, was less likely to disintegrate in a heavy sea than a worm eaten one, and it is revealing that Lloyd’s Register systematically recorded whether and when ships were sheathed, either in wood or copper. Copper sheathing, by keeping the hull clean, also enhanced a ship’s manoeuverability. Between 1786 and 1816 the recorded share of the British merchant fleet coppered rose from 3.3 percent in 1786 to 17.9 in 1816, and by 1830-31 this proportion had risen to 36 per cent.12 A second innovation in shipbuilding during our period was the increasing use of wrought iron joints, and later bracing, to strengthen the ships’ hulls. This was part of the increasing use of iron in shipbuilding, prompted in part by shortages of wood during the wars and in part by technological progress in the making and shaping of iron. For example, Sichel (2017) has shown that the real price of nails fell during the late eighteenth and early nineteenth centuries as iron became cheaper and nails were cut rather than forged. But our concern here is with safety. A ship’s hull can be viewed as a hollow beam whose top is the deck. A weak deck insecurely attached to the hull results in a flimsy vessel that will flex markedly in heavy seas, causing the ship to leak dangerously and, in sufficiently bad conditions, to snap its masts and sometimes to break apart. Traditionally, ships’ decks were pegged to their hulls using expensive wooden brackets called knees, made from the forks of oak trees.

12 This is based on a sample of more than 3,500 entries for the letters A to C in Lloyd’s Register of Shipping 1831. Even with the advent of steamships, the fact that coppered hulls did not foul with weed like iron ones, which needed frequent careening, gave sailing ships a speed advantage on longer voyages (Graham, 1956).

16 Heavy wartime building of naval vessels led to severe shortages, of knees big enough for large ships especially, at a time when puddling was increasing the supply of cheap, high quality wrought iron (Goodwin 1997). The replacement of wooden knees with iron ones secured by bolts was pioneered by the Surveyor (chief architect) of the East India Company, Gabriel Snodgrass, who also began to introduce iron bracing between a ship’s ribs, which stiffened the hull like the diagonal bar on a wooden gate. As a result, expensive oak brackets weakly attached to hulls using wooden pegs were replaced with wrought iron ones fastened with iron bolts. The use of iron later extended to bracing the interior of hulls, greatly increasing their stiffness. In other words, the transition from wooden to iron ships was not a sudden jump but a gradual evolution. Albion (1926: 394) observed that “Practically all the innovations in British naval architecture during this period were occasioned by the inadequate supply of timber.” These innovations were gradually adopted by the navy, and by 1801 naval dockyards were using 1,400 tons of iron annually even before the general adoption of iron knees in 1805 (Lambert 1991: 60–64; Goodwin 1997). How rapidly did iron reinforcement spread to merchant shipping? Our information on this question, which requires cautious handling, comes from Lloyd’s Register, which recorded whether vessels had iron supports from 1816 until 1833, and from newspaper advertisements. Insofar as knees enhanced safety at sea, it was in the interest of both ship-owners and the publishers of Lloyd’s Register to have knees noted; the former because inclusion could have entailed a reduced premium and the latter because it was useful information for insurers. But Lloyd’s Register’s decision to cease publishing the presence of iron knees after its reorganization in 1833- 34 might be interpreted in different ways. Perhaps the new format reflected the belief that the information on knees was no longer useful because they were already universal or nearly so—implying that they were seriously under-recorded before the reorganization. But the proportion of ships advertised in the commercial press as containing iron knees in the 1830s was still modest, which could mean either that both sources were relatively accurate or that both were under-recording by the same proportion. We do know, from post-1833 surveyors’ reports, that the material from which knees were made was still being noted and that some ships built after 1830 were still being fitted with wood knees.13

13 Based on examination of the 44 survey reports done between 1834 and 1852 that have been posted on the Lloyd’s Register Foundation Heritage & Education Centre site

17 As early as 1805 eight per cent of ships of 200 or more tons that were offered for sale in London’s Public Register had iron knees; in 1820 this had risen to ten per cent. But in Lloyd’s Register, the proportion of ships of two hundred tons or more that were recorded as having iron knees was only 1-2 per cent in 1818. It subsequently rose to 16 per cent in 1824 and about 35 per cent in 1831-32.14 This would seem to imply that significant gains in safety from this source were confined to the end of our period, which runs against the evidence from newspaper advertisements in the 1800s and 1810s.15 The entry in Reese’s Cyclopedia [1819] entry on “Shipbuilding” also suggests more general adoption: “Wooden knees having become scarce for some years past, many substitutes have been attempted; and iron knees...are certainly best when properly applied.” Thus, for whatever reason, Lloyd’s inspectors would seem to have under-recorded the presence of knees in the late 1810s. By 1830 Hedderwick’s Treatise on Marine Architecture takes it for granted that all fastenings on the larger ships that are his concern are made of iron. This would seem to contradict the evidence of both LR and the Public Ledger. But Hedderwick was concerned with best practice, which would mean that the flow of new ships was more likely to contain iron knees than the stock recorded in the Registers. It might be thought that with the number of ships recorded being more or less static, the scope for new ships to change the stock would be limited, but that is contradicted by evidence from Lloyd’s Register. Of the ships with names beginning with

(https://hec.lrfoundation.org.uk/archive-library/ships). Unfortunately, no original surveys survive for the period up to 1833. 14 This and the results for 1830 later in the paragraph are based on counts of ships in LR, with names beginning A to C. We included those in the categories: hanging knees, iron hanging knees, iron knees, iron standards, iron standards and knees. Iron knees and iron hanging knees dominated. 15 In a search for references to ‘iron knees’ in the commercial press the number of references involving the sale of ships with knees identified rose from 16 in the 1800s to 49 in the 1810s, 71 in the 1820s, and 117 in the 1830s, while the average age of ships with knees rose from 1.4 years in the 1800s to 6.8 years in the 1830s [based on a search at britishnewspaperarchive.co.uk.]. Given that the number of sailing ships registered remained static over this period, this indicates a somewhat earlier diffusion of iron knees than that implied by LR. The overwhelming majority of the sales reported were in the London Public Ledger & Daily Advertiser, with Gore’s Liverpool General Advertiser coming a distant second. Note, however, that to argue that significant under- recording persisted ignores the presumptions that both LR’s surveyors and ship-owners should have been keen to have knees noted; the former because they enhanced safety and the latter because inclusion would have entailed a reduced premium.

18 the letter “A” present in the 1800 volume, only 29 per cent were still there in 1810. For the 1810 and 1820 volumes the ten-year survival rates were 26 per cent and 41 per cent. Two factors make for these quite low survival rates. The first is simply ships losses during an entire decade, which had clearly been higher during the French wars. The other is brief appearances and disappearances of foreign ships in the Registers. The high turnover means that it is possible for the stock to change its composition quite rapidly. Older ships were also being re-fitted with iron knees. While a more detailed examination of Lloyd’s Register for 1830 shows that about 20 per cent of ships of 200 tons or more built before 1815 had iron structural elements, inspection of earlier editions of the Registers indicates that for two-thirds of these older ships’ wooden elements had been replaced by iron elements during the 1820s. For large ships built from 1815 onward the share with iron knees and standards rose during the late 1810s and 1820s to reach 30 per cent or so, still apparently far from complete adoption. The 1830 Lloyd’s Register also confirms that the use of iron supports increased with the tonnage of the ship. Less than a tenth of ships less than 200 tons had iron features; over a fifth of ships with 200-299 tons and almost a third of ships of 300 tons and more did. Moreover, iron structural elements were more common on vessels destined for distant ports and on vessels destined for icier regions, although this may only reflect the prevalence of larger vessels in these trades. A cautious reading of the evidence might be that iron knees were already being adopted in merchant shipping from the 1800s and that they were common but by no means prevalent on bigger ships by the 1830s.16 Besides iron bracing, Snodgrass’s other major innovation—possibly influenced by his contact with Indian vessels—was a ship with a single, convex flush deck that could be made watertight by battening down hatches, and was far sturdier in heavy seas than traditional European

16 In the 1874 rules for classification (LR 1874, p. 34): “Ships which proceed to sea without being fastened with the iron knees and riders prescribed by the Rules, will have one year deducted from the period to which they would otherwise be entitled to be classed in the Register Book.” (with the footnote specifying that this applies to all wooden vessels). This suggests that even in the 1870s iron knees were standard, but not universal.

19 designs with stepped decks (Snodgrass 1797; Parkinson 1937: 135–138).17 The Royal Navy stuck entirely with deep-waisted ships until the controversial ‘Symonds Frigates’ of the 1830s (Leggett, 2015, 26–58); and merchant shipyards were equally reluctant to adopt flush decks: discussing the inferiority of deep-waisted ships in the 1820s James (1822, 18) noted that “the generality of merchant vessels are, to this day, built in that manner.” They became a well-known feature of the ‘Blackwall Frigates’ of the late 1830s which, as their name suggests, were strongly influenced by the Symonds Frigates that were built in the same yard (Lubbock 1922: 131– 137) and pre-figure the clipper ships of the next decade. However, flush decks had begun to spread beyond the East India Company much earlier.18 As early as 1805 the Public Ledger contained advertisements for about a hundred ships so equipped, with the earliest date of build being 1799. The share of ships of 200 tons or more advertised with flush decks rose from 14 per cent in 1805 to 21 per cent in 1820 (Public Ledger 1805, 1810, 1820). Many of the advertisements indicate that many ships with flush decks were used in the West Indies trade. An even simpler innovation pioneered by Snodgrass was to introduce partitions inside hulls to stop ballast shifting in storms and capsizing the vessel.19 Although a biography of naval architect Sir Samuel Bentham, who introduced watertight bulkheads as a safety measure to the Navy in 1795, states that the innovation spread ‘at length’ to private dockyards (Bentham 1862: 120), we have no way of tracing the diffusion of this technique in the age of sail. Finally, the growing use of steam power may well have had an indirect effect on ship losses. By the mid-1820s many ports had small steamers in use as ferries or tugs. Lloyd’s List reports many incidents in which these local steamers were used to refloat stranded ships. By the mid-1820s copper sheathing was widely diffused, and a growing number of vessels, particularly in long-distance trades, had iron knees, flush

17 In his treatise on ship design Hutchinson (1791: 42) writes that a flush deck strengthens a hull “like a string to a bow” making EIC ships the “best and compleatest merchant’s ships in the world.” 18 A listing of ships associated with the port of Bristol c. 1800-1838 includes six with flush decks constructed between 1809 and 1818 (Farr 1950). They were built (or, in one case, rebuilt) in Bristol (3), in Jarrow (1), Whitby (1), and Chepstow (1). 19Snodgrass’s innovations are catalogued systematically in his list of recommendations to improve the ships of Royal Navy (Snodgrass 1797).

20 decks and watertight bulkheads. These innovations in shipbuilding should have contributed to the reduction in ship losses, though we cannot put a figure on their impact, individually or collectively.

Technological Innovation in Navigation Safe navigation requires reliable charts, compasses, and the means to determine longitude and latitude, and all of these improved to varying degrees during the late eighteenth and early nineteenth centuries. Here we review both the innovations and the extent to which they were put into practice to improve safety and thus had the potential to explain falling loss rates. Some required seamen with sufficient knowledge to apply the tools. Others proved too expensive for widespread application. It was known since the middle ages that the latitude of a ship can be calculated approximately by the altitude of the Pole Star above the horizon or, as the Portuguese learned when sailing south towards the equator in the fifteenth century, the height of the noonday sun. Traditionally, sailors used astrolabes or staffs to measure the height of the sun or a star, a difficult exercise on a rolling deck. Systematic readings only became possible with John Hadley’s octant from 1730, which worked by moving the reflection of the sun or a star down until it lined up with the horizon, allowing the angle between them to be read accurately (Cotter 1968: 57–91). Lighter and far more accurate instruments, capable of measuring wider angles, became practical through one of the most important innovations of the early Industrial Revolution, Ramsden’s 1775 dividing engine, which allowed compact and precisely cut graduated scales to be mass produced at low cost. It is reckoned that by the time of his death in 1801 Ramsden had produced 1450 sextants, while his dividing engine was also availed of by other sextant manufacturers (Baker 2016; Dunn and Higgett 2014: 175; McConnell 2007: ch. 3; Mörzer Bruyns and Dunn 2009). Longitude Longitude can be calculated using the difference between the time in some reference port and the ship’s local time. Local time can be calculated once latitude has been measured, and it was known from the early sixteenth century that reference time can be measured in two ways: either mechanically by a clock that tells the time in the home port; or astronomically by the position of the

21 moon against the background of fixed stars.20 The quest for an accurate measure of longitude and safety at sea were closely linked; in the words of the Longitude Act of 1714, ‘…nothing is so much wanted and desired at sea, as the discovery of the longitude, for the safety and quickness of voyages, the preservation of ships, and the lives of men…’ As noted earlier, it is still widely believed that Harrison’s 1759 H4 sea watch solved the problem of longitude. Harrison’s design differed radically from what other watchmakers of the time would have produced, in particular in its use of a large, heavy balance wheel, which meant that the watch was not self- starting. It was complex and expensive: but it was also delicate and unreliable. Starting in the late 1750s, Pierre Le Roy (credited by Gould (1923, 86) as the effective inventor of a practical timekeeper for navigation) instead designed a chronometer from first principles. Further improvements by John Arnold (whose chronometers were bought in substantial numbers by the East India Company), and Thomas Earnshaw led by around 1810 to a design that changed little until chronometers became obsolete in the second half of the twentieth century. For a half century or more after the invention of H4 chronometers, among the most complicated artefacts of their time, were too expensive and, more importantly, too unreliable for widespread adoption (Dunn and Higgitt 2014: 104–125). Of the 22 chronometers brought on the circumnavigation of HMS Beagle of 1831–1836 only 11 still worked at the end of the voyage (another four were left with a surveying expedition) despite being kept in a special cabin and having a professional instrument maker on board to maintain them (FitzRoy 1839: 325– 331). The accuracy of a chronometer not only changed with variations in temperature, humidity, barometric pressure (making surviving chronometer logs a useful source for climatologists), metal fatigue and the quality of lubricating oils, but with the way it was wound: the exquisite care needed in winding

20 Abortive efforts were also made to estimate longitude by variations in magnetic deviation. Another astronomical timekeeper, that Galileo attempted to market as soon as he discovered them, is the position of Jupiter’s moons. The need for a large telescope, despite repeated efforts to develop stabilized marine chairs, made this impractical at sea but observing these satellites became a standard means for map surveyors on land to estimate longitude precisely. Moreover, by failing to account for the gravitational interaction between the moons Galileo’s tables were inaccurate; and it was while observing Jupiter’s moons at the Paris observatory in 1676 to make usable longitude tables that Rømer made the fundamental discovery that light has a finite velocity.

22 chronometers remained a constant anxiety, as shown by the standard manual of Shadwell (1855).21 Although issued to the Royal Navy in limited numbers from the 1790s, only 7 per cent of British warships had a chronometer in 1802 (Rodger 2004: 382–383). Chronometers were rapidly adopted on the East India Company fleet: for a captain expecting to earn £5,000-£10,000 from his personal cargo allowance, paying £65-£105 each for three chronometers was a minor consideration and most Company ships employed them by 1790 (Davidson 2019). However, few merchant ships used them before the mid-nineteenth century (Davidson 2019; Cotter 1968: 29; Glover 2017), trusting instead the traditional way of “running down the latitude”: sailing directly north or south until they reached the latitude of their destination, and then sailing due east or west until they reached it. For example, although the first Hudson Bay Company ship to carry a chronometer sailed in 1817, they were used to calculate longitude regularly only from 1834 (Glover 2017). One more limitation of chronometers was that the necessary estimate of local time required an exact calculation of the ship’s latitude. Overcast weather made this impossible until Sumner devised the method of position-line navigation in 1837, which remained the basis of British navigation for the next century. The other approach to longitude estimation at sea, that of lunar distances, uses the fact that the relatively rapid movement of the moon across the sky allows it to function like a minute hand against the clock dial of fixed stars. This means that with appropriate tables the angle between the edge of the moon and a known fixed star can be used to calculate the time in the reference port. So, for example, if on July 27 1809 the adjusted angle between the edge of the Moon and Antares was 67°13’3”, after looking up the Nautical Almanac for that day the navigator knew that the time at Greenwich was 18 minutes and 39 seconds after midnight. The Paris Observatory was founded in 1667 for the explicit purpose of obtaining an accurate star map for lunar navigation, as in 1675 was the London Royal Observatory (for “rectifying the tables of the motions of the heavens . . . so as to find out the so much desired longitude of places for the perfecting the art of navigation”). However, because the moon is affected by the sun’s gravity as well as the earth’s, modelling its path accurately enough for reliable navigational tables led to a challenging three body problem that defeated the geometrical approach of Newton (Wepster 2009: 8–25) and whose eventual solution led to an unedifying

21 On the difficulties of maintaining early precision instruments on the move see Baker (2012).

23 priority dispute among Clairaut, d’Alembert, and Euler (Bodenmann: 2010). It was only in 1755 that the German astronomer Tobias Mayer, developing equations devised by Euler to solve the interaction between the orbits of Jupiter and Saturn and effectively solving a least squares problem (Stigler 1986: 11–61), computed tables accurate enough to predict longitude to one degree. In 1806 Johann Karl Burckhardt, using the refined lunar equations of Laplace, devised tables about 12 times more accurate. At the same time that the Board of Longitude finally awarded Harrison £10,000 for his watch, it also gave £3,000 to Mayer’s widow, and £300 to Euler. The practical difficulty in applying lunars lay in “clearing” the observed angle of the effects of refraction, parallax and horizon dip in order to calculate the true angle: a non-trivial problem in spherical trigonometry whose most elegant solution was devised by Borda in 1778 (Gascoigne 2015). Although navigation manuals provided worked examples of lunar estimates that take only about one third of a page to calculate, these may have been beyond the capabilities of most captains, although the exception again is the EIC. In 1768 all new officers were required to be able to use lunar distances and during the 1770s they were in use on half of EIC voyages, although only in the vicinity of known danger points (Miller 2015; Davidson 2016, 2019).22 In summary then, longitude estimation would have had little impact on the safety of ordinary merchant ships before the end of our period.

Charts Although precise longitude estimation may have been beyond ordinary navigators, it was indispensable for making accurate charts. A fundamental problem for navigation was the crude state of hydrographical knowledge: the standard book of charts of the British coast through most of the eighteenth century was Grenville Collins’ rudimentary Great Britain’s Coasting Pilot (first published in 1693 and frequently republished, reaching its twenty-first edition in 1792), along with somewhat better French and Dutch charts. Although the Royal Navy had supported the surveying work of James Cook and others in the 1760s and 1770s, it established a hydrographic department only in 1795, and did not sell its maps until 1821. Similarly, the East India Company’s hydrographer, Alexander Dalrymple,

22 This runs counter to the claim of Wess (2016) that log book entries in EIC and Royal Navy ships before 1800 report only dead reckoning. John Brisbane, astronomer on an EIC ship bound for Paramatta in New South Wales, recalled in 1795 how ‘in that immense fleet there was perhaps not 10 individuals who could make a lunar observation’ (as cited by Phillips 2016: 40) but by that stage chronometers were in extensive use on EIC ships.

24 produced large-scale maps based on novel surveying techniques for Company use, but not charts for use at sea (Fisher 1991, 2011: 60). The captains of the Hudson Bay Company also developed effective charts for navigating the waters of the Arctic and the Bay (Glover 2017). However, in the late eighteenth century privately produced and crowd- sourced charts began to appear; these were known as Bluebacks from the colour of their heavy backing paper. The first and most important was the large chart of the English Channel by John Hamilton Moore who estimated that it had sold “upwards of 5,000 copies” between its first appearance in 1786 and 1792 (Petto 2015: 79– 122). Each chart was sold with a detailed pilotage manual (such as Dessiou 1802) that for each port gave times of high water, depth soundings, and guides to beacons and channel marking buoys (themselves indicative of direct government efforts to make approaches to ports safer). Moore also produced charts of the Mediterranean, the Baltic, the east coast of America, and the West Indies that hardly differ from their modern counterparts, giving longitude and latitude, precise outlines of the coast with insets for major harbours, depth soundings, and descriptions of the sea bottom. Although Admiralty charts were sold at considerably lower prices, Bluebacks, by then mostly printed by John Norie, remained the choice of most ships’ masters until well into the nineteenth century, coming as they did with detailed pilotage manuals, and being designed to be legible in dim candlelit cabins at sea, with navigational hazards clearly highlighted (Fisher 2003, 2011). The quality of charts increased significantly with the introduction of the station pointer in the 1770s (Fisher 1991: 121). Because most ship losses took place in coastal waters, better charts are likely to have been a primary factor in reducing loss rates.23

[Figures 4 and 5 about here]

Navigation Manuals These navigational innovations mattered little if mariners lacked the skills to apply them. Although state-run navigational schools in continental Europe dated from the time of Prince Henry the Navigator, Britain characteristically relied on informal education. Private tutors were numerous from Elizabethan times but the

23 An additional factor may have been increased density of shipping. For a vessel approaching land, the simplest and most reliable way to determine its position was to hail a coaster or outbound ship for its exact location. The probability of being warned away from danger increases at a linear rate with the amount of traffic.

25 earliest systematic navigation textbook was John Robertson’s 1754 Elements of Navigation, though its uncompromising reliance on spherical trigonometry made it incomprehensible to most sea captains. The first useful manual, priced at ten shillings (roughly the weekly wage of a seaman in the 1770s: Smith 1776, I: 135) and largely based on worked examples, was again due to John Hamilton Moore. His New Practical Navigator of 1772 started with arithmetic and elementary trigonometry before taking the reader successively through use of compass and log line; plotting course on charts with plane, traverse, mid-latitude and Mercator sailing; estimating tides; recording hourly course and speed on a traverse board; calculating local time and latitude; and finally calculating longitude using lunars. At the end were tables of log trigonometric functions, refraction, parallax, the sun’s declination, and the right ascension of the sun and major stars. Moore’s structure was kept in successive editions of the two most widely used manuals: Norie’s New and Complete Epitome of Practical Navigation which first appeared in 1805; and its American equivalent, Nathaniel Bowditch’s American Practical Navigator (which began as a pirated edition of Moore; see Cotter 1977) from 1802 onward. However, despite covering advanced techniques, in Moore’s (1794: 186) view “the most capital part of navigation” for the young mariner was the systematic working up of a daily journal of position. This started from the traverse board of hourly speed and heading, making corrections for compass deviation and leeway, and estimating position on a chart using mid-latitude sailing. Successive editions of these manuals provide a useful way to track progress in navigational practice. The early editions of Norie are almost identical to Moore, although the exposition in general is more lucid and the algorithms for calculating lunars are considerably simpler. By 1835, however, Norie describes how to adjust for the compass deviation caused by the growing amount of iron on ships; and, instead of chronometers being placed as a short appendix after lunars, they are now discussed at length before lunars appear. Clearly the mariners who read these manuals had be both literate and numerate (Schotte 2019). Although it seems likely that educational standards of officers rose in response to the increased complexity of applied navigation, formal examinations to certify navigators and masters only became widespread in the 1850s (Vasey 1980). Early navigation manuals were plagued by inaccurate tables of logarithms and trigonometric functions. The most ambitious effort to produce reliable tables, intended for a cadastral survey of post-revolutionary France, was undertaken by de Pronys in 1794 with algorithms designed by Legendre and others. Inspired by Adam Smith’s discussion of division of labour, he established a “computation factory” of unemployed hairdressers—accustomed to painstaking work, but

26 victims of the reaction against the elaborate coiffure of Bourbon times—to perform the routine calculations. However, the completed tables could not be printed because of the collapse in value of the assignat (Grattan-Guinness 2003). Another ambitious but abortive project to develop mathematical tables for navigation was Charles Babbage’s 1822 idea of a Difference Engine (Swade 2001). Although longitude estimation contributed less to safety at sea in our period than usually thought, that does not mean that progress in navigation was lacking. Most ships still navigated by dead reckoning based on speed and compass heading until the 1830s but, thanks to Moore and his imitators, it was a far more sophisticated and reliable dead reckoning than the crude guesswork of the 1770s.

Compasses In contrast to the progress in positional estimation and chart making, the improvement of the oldest and most important navigational instrument, the compass, was remarkably slow. Small compass errors translate into large, and potentially deadly, errors in estimated position: heading ten miles due west on a compass bearing that is only 6 degrees in error will leave a ship one mile north of its estimated position. Three difficulties plagued compasses: low quality iron, magnetic variation, and magnetic deviation. In 1745 the English physicist and inventor Gowin Knight devised a process to magnetize steel bars resulting in a compass needle that retained its magnetism longer than soft iron ones, and this technique became public after his death in 1776. Magnetic deviations were strongest in high latitudes with their strong magnetic fields. Despite considerable efforts to improve compasses, the Ross Arctic voyage of 1818, which was intended in part to assess the performance of novel designs, found all of them to be extremely unreliable, pointing in widely different directions (Dunn 2016). Magnetic variation—the difference between magnetic and true north—had been familiar since Elizabethan times (and seemed at first to offer a way to measure longitude, being first systematically mapped by Halley on his voyage around the Atlantic in 1701), but it was frequently ignored. The navigators of Shovell’s fleet, wrecked on the Scilly Isles in 1707 with the loss of 2,000 lives (the event that led directly to the establishment of the Board of Longitude), had not compensated for a 10-degree variation, as well as relying on charts that placed the islands nine miles north of their actual position. Such a disaster would have been prevented a century later, when charts were much more reliable and Norie and Bowditch, in their widely-consulted manuals, showed that magnetic variation may

27 easily be compensated for by comparing the compass position of the sun at dawn with the true position in published tables. An untoward consequence of the increasing use of iron reinforcing and cables after 1800 was that it worsened the deviation of compasses from magnetic north. However, manuals described how to compensate by comparing the position of a compass needle when the ship was heading east-west with its position heading north-south. Placing the compasses high above deck was also found to help (Quinn 2001).24 While compass design was largely stagnant, notable improvements occurred during our period in the two other traditional mainstays of navigation: mechanical log lines (for estimating speed), particularly Edward Massey’s design of 1802 (Turner 1998: 35); and rapid depth sounding, although the latter was only needed when fast steamships appeared. Hand-held spyglasses, widely used from the seventeenth century to identify navigational hazards and safe places to land, considerably improved during the eighteenth century. In 1758 John Dollond patented a lens which corrected for chromatic aberration, and joined a partnership formed by his son to sell spy-glasses incorporating the new lens. With the termination of Dollond’s patent in 1782, cheaper achromatic telescopes became widely available (Dunn 2011: 73–76).

State Involvement The second half of the eighteenth century also saw increased state efforts to improve navigational aids around coasts. The number of lighthouses on the east coast of the United States rose from three (all in Massachusetts) before 1750, to about twenty-four by 1800, and eighty-five by 1830. In the United Kingdom the numbers rose from about fifteen in the mid-eighteenth century to 57 by 1800, 118 by 1830, and 264 in 1844.25 Steady innovation occurred in light-houses, detailed chronologically by Stevenson (1959: 61–85): in particular the replacement of simple coal fires and candles with oil Argand Lamps illuminating parabolic reflectors; and the building of “wave-swept” lighthouses off shore, pioneered by Smeaton’s 1759 Eddystone Lighthouse. Smeaton’s sturdy stone construction served as a model for many others, with innovations that included keeping the centre of gravity low, using a type of concrete that solidified under water, and

24 After the period that concerns us, on iron clippers and steamships compasses were useless (as illustrated by the 1854 sinking of the clipper RMS Tayleur off the east coast of Ireland with 370 drowned). Although adjustable magnets began to be tested in the 1850s, the problem was not solved until Lord Kelvin’s binnacle design in the 1870s. 25 Probert (1999). Some of these numbers were compiled from Wikipedia entries.

28 ensuring durability with a system of interlocking granite blocks.26 Talented innovators like John Rennie and Robert Stevenson, surveyor to the Commissioners of Northern Lights, built on Smeaton’s achievements. At the same time channel- marking buoys and beacons were increasingly installed by local harbour commissioners. The first successful purpose-built lifeboat was designed by Henry Greathead, a boat-builder at South Shields, in 1789. Built to carry twenty people, it could be rowed in either direction and rose at both ends to reduce the risk of foundering in heavy seas. Other hallmarks included a curved keel, short oars, cork casing for buoyancy, copper plates for rigidity, and the absence of a rudder. Versions of Greathead’s invention, manned by experienced oarsmen who were usually well rewarded for their efforts, were soon operated by lifeboat societies in several British ports and these societies were organized into what became the Royal National Lifeboat Institution in 1824.27 Manby’s mortar, a primitive form of zip-line, was invented by Captain George Manby in 1808. This invention was a close cousin of the less costly and lighter ‘Rocket’ apparatus invented by Henry Trengrouse, also in 1808. Trengrouse devised his apparatus, which included a chair to carry those rescued ashore, as a reaction to the loss of life caused by the wreck of the Anston in 1807 (Pollard 2004). All stations of the Preventative Water Guard (1809, reformed as the Coast Guard in 1822] were presented with a Manby mortar. Although the main remit of the Water Guard was to prevent smuggling, it was also intended to help in the event of shipwreck. The mortars were credited with saving over a thousand lives by mid-century (Prosser 1885). Innovations such those of Greathead, Manby, and Trengrouse were part of a rising humanitarian concern at loss of life at sea that drove campaigns for state intervention, reflected in the foundation in London in 1774 of the Society for the Recovery of Persons Apparently Drowned (the Royal Humane Society from 1787) and the passage of the Burial of Drowned Persons Act in 1808.28 A belief that the risk of shipwreck was increasing—blamed on “bad vessels, bad navigation and drunken officers . . . in more or less equal parts” (MacDonagh 1961: 48)—seemed to be confirmed by McCulloch’s (1835) influential article on shipwrecks in the

26 Anon. (1844) remains a useful survey of improvements in lighthouse technology. 27 For example, Dublin’s port authority placed five lifeboats at locations around Dublin Bay between 1801 and 1816 (Gilligan 1980). 28 Note that our concern has been with mortality associated with shipwrecks; mortality on board ship, mainly from infectious diseases, did not witness any improvement between 1820 and 1860 (Cohn 2009: 142-54).

29 Edinburgh Review, and the detailed accounts of individual incidents published in Alexander Becher’s Nautical Magazine.29 In conclusion, what we can see is a pattern of sustained innovation and the gradual adoption of improved technology and techniques both in navigation and shipbuilding. None of these technical changes is the “silver bullet” that in itself would explain the downward movement in loss and insurance rates that we observe in the data. Experience on the long and hazardous route to India provides a way of looking at their collective impact. As our discussion of technological change has shown, the East India Company was pioneer both in ship construction and in navigation. By the 1820s its loss rate had fallen to 0.6 per cent, less than half the rates on the much shorter trans-Atlantic routes and far less than the 3.1 per cent loss rate for non-Company ships on the Indian service. The fall in loss rates on other routes from the 1780s to the 1820s can be seen as movements toward the East India Company standard.

Conclusions: The safety revolution in oceanic shipping that we have outlined here ties in closely with wider debates on the nature of the Industrial Revolution. For the last generation, following the work of Crafts and Harley (1992), reiterated by Crafts (2014), the Industrial Revolution has become almost synonymous with advances in cotton, iron, and steam. While it is unquestionable that in macroeconomic terms these sectors dominated growth, this focus on cotton, iron and steam has tended to distract attention from the important advances taking place in other sectors such as brewing, pottery, glass, paper, printing, hydraulics, mechanical engineering, and non-ferrous metals, as emphasized by Mokyr (2009, 131–144), Berg and Hudson (1992) and Allen (2014); see also the more recent studies of rapid progress in gas lighting by Tomory (2012) and watchmaking by Kelly and Ó Gráda (2016).30

29 The Nautical Magazine was founded in 1832. Its account in 1838 of cannibalism on board the brig Caledonia was one factor leading to the creation in 1839 of the Select Committee on Shipwrecks of Timber Ships. Probert (1999: 78fn) describes McCulloch’s article as “an authoritative plea from politically motivated sources”. Compare Vasey (1980: 4). 30 To take watchmaking as an example, assuming that Britain produced 200,000 watches worth an average of £1 c. 1800 and that British national income c. 1800 was £200-250 million would imply that watches then contributed at most 0.1 per cent to national income (Kelly and Ó Gráda 2016: 1730). That means that the macroeconomic impact of productivity change in the sector was small whereas its technological spillovers were considerable, especially in developing the early textile machinery of Manchester as Musson and Robinson (1969: 430-31, 437-39) demonstrated.

30 Complementing our interest in transportation at sea, Bogart (2014) has demonstrated the large advances made on land at this time. The improvements in shipping safety saved lives—we can’t say how many—but may not have constituted a huge reduction in shipping costs. Ships were major capital expenditures and capital costs probably accounted for less than a third of total shipping costs. For a ship financed at 5 per cent and amortized over 15 years, savings of one per cent on marine insurance for two voyages per year might amount to around 5 per cent of total costs. If the reduction in losses prolonged average ship lives, the gains would have been somewhat greater. But the important point is that these improvements in safety are just one of several indicators of progress in sailing ship technology. Copper sheathing reduced capital costs and increased ship speeds, which in the case of the slave trade reduced mortality on the Middle Passage and brought slaves to the Americas in better condition (Solar 2013; Solar and Rönnbäck 2014; Solar and Hens 2015; Solar and Duquette 2017). Sailing ships on many routes became larger and required fewer crew members per ton (Lucassen and Unger 2000). Shipping during the early Industrial Revolution does not appear to have been the technologically stagnant sector suggested by the work of North (1868); indeed, the ‘substantial productivity advance’ in shipping—0.6 per cent annually in the first half of the century— identified by Harley (1988, 861) surely owes something to the improvements documented above. Moreover, this is an indicator of productivity change in the service sector, which is poorly measured in existing national output estimates. Our results also bear directly on Mokyr’s concept of an Industrial Enlightenment (Mokyr 2016) in which European science, driven by an ideal of creating useful knowledge, made major contributions to the development of technology. Although not mentioned by Mokyr, the earliest and most direct example of this Industrial Enlightenment at work is the quest dating from the mid- seventeenth century to improve astronomical knowledge in order to advance navigation. The Paris and Greenwich Observatories were established for the stated purpose of providing astronomical data for reliable navigation tables; and many of the major figures of seventeenth and eighteenth century science—including Galileo, Newton, Hooke, Huygens, Euler, Rømer, and Laplace—were directly engaged in improving navigation. In fact, Wepster (2009, 13) argues that the annual Académie des Sciences essay prize, which alternated between a topic in general astronomy and one in navigation, played just as important a role in advancing navigation as the rewards offered under the British Longitude Act. The improvements in navigation also illustrate the important complementary input of artisan skills. In fact, most of the innovations that came to dominate navigational practice by the 1840s were due to ordinary watchmakers and seamen.

31 The two most important early contributions to practical navigation—accurate charts and accessible navigational textbooks—were both pioneered by the retired mariner turned navigational instructor John Hamilton Moore. Even for longitude estimation, the complicated lunar distances of astronomers (disparaged by Harrison as “professors” and “priests”: Gould, 1923: 68) were eventually superseded by new methods (notably position line navigation) that combined artisanal chronometers, astronomical observation and dead reckoning. The production of affordable, accurate sextants was made possible by instrument maker Jesse Ramsden’s 1771 dividing engine, an invention which underlay the development of all subsequent measuring instruments. By 1789 Ramsden had produced a thousand sextants, and his London factory, which employed sixty workers, produced many more (Baker 2016; Dunn and Higgett 2014: 175). Just as Lancashire watchmakers provided the technical expertise to build the earliest textile machinery, accurate measurement in turn made possible the development in Manchester in the 1820s of heavy machine tools without which the tens and later hundreds of thousands of iron mules and power looms that lined Victorian cotton mills could not have been built.

References

Albion, Robert Greenhalgh. Forests and Sea Power. Cambridge: Harvard University Press, 1926.

Allen, Robert C. “Technology.” In The Cambridge Economic History of Britain, Vol. 1, edited by Roderick Floud, Jane Humphries, and Paul Johnson, 292-320. Cambridge: Cambridge University Press, 2014.

Anon. Smeaton and Lighthouses: A Popular Biography, with an Historical Introduction and Sequel. London: Parker, 1844.

Baker, Alexi. “‘Precision,’ ‘Perfection,’ and the Reality of British Scientific Instruments on the Move During the 18th Century.” Material Culture Review 74–75 (2012): 14–29.

Berg, Maxine and Pat Hudson. “Rehabilitating the Industrial Revolution.” Economic History Review 45, no. 1 (1992): 24–50.

Bentham, M. S. The Life of Brigadier Sir Samuel Bentham … by His Widow. London: Longman Green, 1862.

Bodenmann, Siegfried. “The 18th Century Battle over Lunar Motion.” Physics Today 63 (2010): 27–32.

32 Bogart Dan. “The Transport Revolution in Industrializing Britain.” In The Cambridge Economic History of Britain 1700 to 1870, Vol. 1, edited by Roderick Floud, Jane Humphries and Paul Johnson, 368-391. Cambridge University Press, Cambridge, 2014.

Chet, Guy. The Ocean is a Wilderness: Atlantic Piracy and the Limits of State Authority, 1688- 1856. Amherst: University of Massachusetts Press, 2014.

Clark, Geoffrey. “Insurance as an Instrument of War in the 18th Century.” The Geneva Papers on Risk and Insurance 29 (2004): 247–257.

Cockerell, H.A.L., and Edwin Green. The British Insurance Business: A Guide to its History & Records. Sheffield: Sheffield Academic Press, 1994.

Cohn, Raymond L. Mass Migration Under Sail: European Immigration to the Antebellum United States. Cambridge: Cambridge University Press, 2009.

Cook, E.R., R.D. D’Arrigo and M.E. Mann. 2002. “A Well-Verified, Multiproxy Reconstruction of the Winter North Atlantic Oscillation Index since A.D. 1400.” Journal of Climate 15 (2002): 1754–1764.

Cotter, C.H. A History of Nautical Astronomy. London: Hollis and Carter, 1968.

Cotter, C. H. “John Hamilton Moore and Nathaniel Bowditch”. Journal of Navigation 30, no. 2 (1977): 323-26.

Crafts, N. “Productivity Growth during the British Industrial Revolution: Revisionism Revisited”. CAGE Working Paper No. 204, University of Warwick, 2014.

Crafts, N. F. R. and C. K. Harley. “Output Growth and the British Industrial Revolution: A Restatement of the Crafts-Harley View”. Economic History Review, 45, no. 4 (1992): 703-30.

Davidson, Simon C. “The Use of Chronometers to Determine Longitude on East India Company Voyages”, The Mariner's Mirror, 102:3, (2016) 344-348.

Davidson, Simon C. “Marine Chronometers: the Rapid Adoption of New Technology by East India Captains in the Period 1770–1792 on over 580 Voyages,” Antiquarian Horology, 40, no. 1 (2019): 76–91.

Davis, Ralph. The Rise of the English Shipping Industry in the 17th and 18th Centuries. 2nd ed. Newton Abbot: David and Charles, 1972.

Denzel, Markus. “The Price of Minimalizing the Risks at Sea: The Hamburg Marine Insurance Rates in the 18th and Early 19th Century.” In I prezzi delle cose nell’eta preindustriale: selezione di ricerche, edited by Giampiero Nigro, 367-83. Florence: Firenze University Press, 2017.

33 Dessiou, Joseph F. Sailing Directions to be used with John Hamilton Moore’s Chart of the British Channel, and the South-West Coast of Ireland ... Sixth edition, etc. London: John Hamilton Moore, 1802.

Dixon, Conrad Hepworth. “Seamen and the Law: an Examination of the Impact of Legislation on the British Seaman’s Lot, 1588-1918”, PhD thesis, University College London, 1981 [available online at: http://discovery.ucl.ac.uk/1317735/1/282305.pdf].

Dunn, Richard. The Telescope: A Short History. London: Conway, 2011.

Dunn, Richard. “North by Northwest: Experimental Instruments and Instruments of Experiment.” In Geography, Technology and Instruments of Exploration, edited by Fraser MacDonald and Charles W.J. Withers, 57-76. London: Routledge, 2016.

Dunn, Richard and Higgitt, Rebekah. Finding Longitude. London: Collins, Royal Museums Greenwich, 2014.

Earle, Peter. Sailors: English Merchant Seamen, 1650–1775. London: Methuen, 1998.

Edler de Roover, Florence. “Early Examples of Marine Insurance.” Journal of Economic History 5 (1945): 172–200.

Farr, Grahame E. (ed.). Records of Bristol Ships 1800 to 1830. Bristol: Bristol Records Society, 1950.

Fisher, Susanna. “The Origins of the Station Pointer.” International Hydrographic Review, 68, no. 2 (1001): 119-26.

Fisher, Susanna. “‘More Attractive and Useful’: the Popularity of Privately Published Blueback Charts in the Nineteenth Century.” The Cartographic Journal 40 (2003): 79–88.

Fisher, Susanna. The Makers of the Blueback Charts: A History of Laurie Norie and Wilson Ltd. St Ives: Imray Laurie Norie and Wilson, 2011.

FitzRoy, Robert. Narrative of the Surveying Voyages of His Majesty’s Ships Adventure and Beagle, Between the Years 1826 and 1836. Appendix to Volume 2. London: Henry Colburn, 1839.

Gascoigne, John. “Navigating the Pacific from Bougainville to Dumont d’Urville: French Approaches to Determining Longitude, 1766–1840.” In Navigational Enterprises in Europe and its Empires, 1730-1850, edited by Rebekeh Higgitt, Richard Dunn, and Peter Jones, 180- 97. London: Palgrave Macmillan, 2015.

Gilligan, H. A. “Captain William Hutchison and the Early Dublin Bay Lifeboats.” Dublin Historical Record 33 (1980): 42–45.

Glover, William. “Using Longitude on Early Nineteenth Century Voyages to Hudson’s Bay.” The Northern Mariner 27 (2017): 355-72.

34 Goodwin, Peter. “The Influence of Iron in Ship Construction, 1660– 1830.” Mariner’s Mirror 84, no. 1 (1997): 26–40.

Gore’s Liverpool Advertiser, various dates.

Gould, R.T. The Marine Chronometer: Its History and Development. London: J. D. Potter, 1923.

Grattan-Guinness, Ivor. “The Computation Factory: de Prony’s Project for Making Tables in the 1790s.” In The history of mathematical tables: from sumer to spreadsheets, edited by Martin Campbell-Kelly, 104-21. Oxford: Oxford University Press, 2003.

Harley, C. Knick. “The Shift from Sailing Ships to Steamships, 1850-1890: A Study in Technological Change and its Diffusion.” In Essays on a Mature Economy: Britain After 1840, edited by Donald N. McCloskey, 215-34. London: Methuen, 1971.

Harley, C. Knick. “Ocean Freight Rates and Productivity, 1740-1913: The Primacy of Mechanical Invention Reaffirmed.” Journal of Economic History 48, no. 4 (1988): 851–876.

Hutchinson, William. A Treatise Founded Upon Philosophical and Rational Principles: Towards Establishing Fixed Rules, for the Best Form ... of Merchant’s Ships ... and Also the Management of Them ... by Practical Seamanship. London: Thomas Billinge, 1971.

James, William. The naval history of Great Britain from the declaration of war by France in February 1793 to the accession of George IV in January 1820... Vol. 1. London: Baldwin, Craddock and Joy, 1822.

John, A. H. “The London Assurance Company and the Marine Insurance Market of the Eighteenth Century.” Economica 25, no. 98 (1958): 126–141.

Kelly, Morgan, Joel Mokyr, and Cormac Ó Gráda. “Ingenious Mechanics: Artisan Skill and the Industrial Revolution,” UCD Centre for Economic Research Working Paper, 2019.

Kelly, Morgan and Cormac Ó Gráda. “Adam Smith, Watch Prices, and the Industrial Revolution.” Quarterly Journal of Economics 131 (2016): 1727– 1752.

Kelly, Morgan and Cormac Ó Gráda. 2019. “Speed under Sail, 1750-1830.” Economic History Review, 72, no. 2 (2019): 459-480.

Kingston, Christopher. “Governance and Institutional Change in Marine Insurance, 1350–1850.” European Review of Economic History 18, no. 1 (2013):1– 18.

Lambert, Andrew. The Last Sailing Battlefleet: Maintaining Naval Mastery 1815–1850. London: Conway Maritime Press, 1991.

Landes, David S. Revolution in Time: Clocks and the Making of the Modern World. Cambridge, Mass.: Harvard University Press, 1983.

35 Leggett, Don. Shaping the Royal Navy: Technology, Authority and Naval Architecture, C.1830- 1906. Manchester: University of Manchester Press, 2015.

Leonard, A. B. “Underwriting British Trade to India and China, 1780– 1835.” The Historical Journal 55 (2012): 983–1006.

Leonard, Adrian. “Underwriting Marine Warfare: Insurance and Conflict in the Eighteenth Century.” International Journal of Maritime History 25 (2013):173–185.

Leonard, Adrian. “The Origin and Development of London Marine Insurance, 1547-1824.” D.Phil. thesis, University of Cambridge, 2014.

Leonard, Adrian. “The Pricing Revolution in Marine Insurance, 1600–1824.” Unpublished working paper, 2016.

Lloyd’s List, various issues.

Lloyd’s Register, various volumes.

London New Price Current, various issues.

Lubbock, Basil. The Blackwall Frigates. Glasgow: J. Brown and Son, 1922.

Lucassen, J. and R.W Unger. “Labour Productivity in Ocean Shipping, 1450-1875.” International Journal of Maritime History 12(2) (2000): 127-141.

McConnell, Anita. Jesse Ramsden (1735–1800): London's Leading Scientific Instrument Maker. Aldershot: Ashgate, 2007.

MacDonagh, Oliver. A Pattern of Government Growth. London: Mc-Gibbon and Kee, 1961.

MacGregor, John. Commercial Statistics: A Digest of the Productive Resources...Vol. 3. London: Charles Knight and Company, 1847

Martin, Frederick. The History of Lloyd’s and of Marine Insurance in Great Britain. London: Macmillan, 1876.

McCulloch, J.R. “On the Frequency of Shipwrecks.” Edinburgh Review 60 (1835): 338–353.

McCusker, John J. "The Early History of ‘Lloyd's List’." Historical Research 64, no. 155 (1991): 427-431.

Miller, David Philip. “Longitude Networks on Land and Sea: The East India Company and Longitude ‘Measurement in the Wild’, 1770-1840”. In Navigational Enterprises in Europe and Its Empires 1730-1850, edited by Richard Dunn and Rebekah Higgitt, 223-47. Basingtoke: Palgrave Macmillan, 2015.

36 Mokyr, Joel. The Enlightened Economy: An Economic History of Britain 1700-1850. New Haven: Yale University Press, 2009.

Mokyr, Joel. A Culture of Growth: Origins of the Modern Economy. Princeton: Princeton University Press, 2016.

Moore, John Hamilton. The new practical navigator: being an epitome of navigation, containing the different methods of working the lunar observations, and all the requisite tables used with the nautical almanack in determining the latitude and longitude, ... the tenth edition. constructed on a new plan, and illustrated with copper plates. London: John Hamilton Moore, 1794.

Mörzer Bruyns, W. F. J. and Richard Dunn. Sextants at Greenwich: A Catalogue. London: National Maritime Museum, 2009.

Musson, A. E. and E. Robinson. Science and Technology in the Industrial Revolution. Manchester: Manchester University Press, 1969.

Naish, G. P. B. “Ships and Ship-building [c. 1500-c. 1750].” In A History of Technology: The Industrial Revolution c1750–c1850, edited by Charles Singer, E.J. Holmyard, A. R. Hall and Trevor J. Williams, Vol. 4, 471-500. Oxford: Oxford University Press, 1958.

Norie, J.W. A New and Complete Epitome of Practical Navigation. London: author and for William Heather, 1805.

North, Douglass C. “Sources of Productivity Change in Ocean Shipping, 1600-1850.” Journal of Political Economy 76 (1968): 953–970.

Parkinson, C. Northcote. Trade in Eastern Seas 1793–1813. Cambridge: Cambridge University Press, 1937.

Pedley, Mary Sponberg. The Commerce of Cartography: Making and Marketing Maps in Eighteenth-Century France and England. Chicago: University of Chicago Press, 2005.

Pearson, Robin and David Richardson, “Insuring the Atlantic Slave Trade.” Journal of Economic History, 79 (2019), 417-446.

Petto, Christine. Mapping and Charting in Early Modern England and France: Power, Patronage, and Production. Lanham, MD: Lexington Books, 2015.

Phillips, Eoin. “Instrumenting Order: Longitude, Seamen and Astronomers, 1770-1805.” In Geography, Technology and Instruments of Exploration, edited by Fraser MacDonald and Charles W.J. Withers, 37-56. London: Routledge, 2016.

Phillips, Sir Richard. A Million of Facts and Correct Data in the Entire Circle of the Sciences. London: Sherwood, Gilbert, and Piper, 1832.

37 Pollard, A.F, revised by R.C. Cox. “Trengrouse, Henry (1772–1854).” In Oxford Dictionary of National Biography. Oxford: Oxford University Press, 2004 [http://www.oxforddnb.com/view/article/27707, accessed 2 Jan 2017].

Prince’s Price Current, various dates.

Probert, William. “The Humanitarian, Technical and Political Response to Shipwreck in the First Half of the Nineteenth Century: The 1836 Inquiry and its Aftermath.” PhD thesis University of Southampton, 1999.

Prosser, Richard Bissell, revised by R.C. Cox. “Manby, George William.” In Dictionary of National Biography. Oxford: Oxford University Press, 2004 [http://www.oxforddnb.com/view/article/17919, accessed 2 Jan 2017].

Public Register and Daily Advertiser, various dates.

Puttevils, Jeroen and Marc Deloof. “Marketing and Pricing Risk in Marine Insurance in Sixteenth- Century Antwerp.” Journal of Economic History 77 (2017): 796–837.

Quinn Paul. “The Early Development of Magnet Compass Correction.” The Mariner's Mirror 87, no. 3 (2001): 303-315.

Palmer, Sarah. “The Indemnity in the London Marine Insurance Market, 1824-50,” In The historian and the business of insurance, edited by Oliver M. Westall, 74-94. Manchester: Manchester University Press, 1984.

Rediker, Markus. Between the Devil and the Deep Blue Sea: Merchant Seamen, Pirates and the Anglo-American Maritime World, 1700–1750. Cambridge: Cambridge University Press, 1987.

Rees, Gareth. 1971. “Copper Sheathing: An Example of Technological Diffusion in the English Merchant Fleet.” Journal of Transport History 1, no. 2 (1971): 85-94.

Rodger, N. A. M. The Command of the Ocean: A Naval History of Britain, 1649-1815. London: Allen Lane, 2004.

Schotte, Margaret E. Sailing School: Navigating Science and Skill, 1550-1800. Baltimore: St. Johns Hopkins University Press, 2019.

Shadwell, C.F.A. Notes on the Management of Chronometers and the Measurement of Meridian Distances, London: 1855.

Sichel, Daniel. 2017. “The Price of Nails since 1700: Even Simple Products Experienced Large Price Declines.” Unpublished working paper, April 2017.

Smith, Adam. An Inquiry into the Nature and Causes of the Wealth of Nations, Vol. 1. London: W. Strahan, 1776.

38 Snodgrass, Gabriel. 1797. “Letter from Gabriel Snodgrass, esq. to the Right Honorable Henry Dundas ... on the mode of improving the navy of Great Britain.” Printed by order of the Hon. Court of Directors, 1797.

Solar, Peter M. “Opening to the East: Shipping between Europe and Asia, 1770-1830.” Journal of Economic History 73, no. 3 (2013): 625-661.

Solar, Peter M. and Klas Rönnbäck. “Copper Sheathing and the British Slave Trade.” Economic History Review 68, no. 3 (2014): 806-829.

Solar, Peter M. and Luc Hens. “Ship Speeds during the Industrial Revolution: East India Company Ships, 1770-1828.” European Review of Economic History 20, no. 1 (2016): 66-78.

Solar, Peter M. and Nicolas J. Duquette. “Ship Crowding and Slave Morality: Missing Observations or Incorrect Measurement?” Journal of Economic History 77, no. 4 (2017), 1177- 1202.

Spooner, Frank C. Risks at Sea: Amsterdam Insurance and Maritime Europe, 1766-1780. Cambridge: Cambridge University Press, 1983.

Stevenson, D. Alan. The World’s Lighthouses before 1820. Oxford: Oxford University Press, 1959.

Stigler, Stephen M. The History of Statistics: The Measurement of Uncertainty Before 1900. Cambridge, Mass: Belknap Press of Harvard University Press, 1986.

Swade, Doron. The Difference Engine: Charles Babbage and the Quest to Build the First Computer. London: Penguin, 2001.

Temin, Peter. “Two Views of the British Industrial Revolution.” Journal of Economic History 57, no. 1 (1997): 63-82.

Tomory, Leslie. Progressive Enlightenment: The Origins of the Gaslight Industry, 1780–1820. Cambridge, Mass: MIT Press, 2012.

Turner, Gerard L’Estrange. Scientific Instruments, 1500-1900: An Introduction. Berkeley: University of California Press, 1998.

United Kingdom. House of Commons. “Report from the Select Committee on Marine Insurance.” Parliamentary Papers 1810 IV (226).

United Kingdom. “Affairs of the East India Company: Minutes of Evidence, 17 June 1830.” Journal of the House of Lords 62 (1830): 1130–1136.

United Kingdom. House of Commons. “Comparative Account of the Population of Great Britain in the Years 1801, 1811, 1821, and 1831.” Parliamentary Papers 1831 XVIII (348).

39 United Kingdom. House of Commons. “Report from the Select Committee appointed to inquire into the Cause of Shipwrecks.” Parliamentary Papers 1836 XVII (567).

United Kingdom. House of Commons. “Report from the Select Committee on Shipwrecks of Timber Ships.” Parliamentary Papers 1839 IX (217).

Vasey, Thomas Watson Cornforth. “The Emergence of Examinations for British Shipmasters and Mates, 1830–1850”. PhD thesis University of Durham, 1980.

Weiss, Leonard. Watch-making in England 1760-1820. London: Robert Hale, 1982.

Wepster, Steven. Between Theory and Observations: Tobias Mayer’s Explorations of Lunar Motion, 1751-1755. New York: Springer, 2009.

Wess, Jane. “Navigation and Mathematics. A Match Made in the Heavens?” In Navigational Enterprises in Europe and its Empires, 1730–1850, edited by Rebekah Higgitt, Richard Dunn and Peter Jones, 201-22. London: Palgrave Macmillan, 2016.

Williams, David M. “Merchanting in the First Half of the Nineteenth Century: The Liverpool Timber Trade.” In Merchants and Mariners: Selected Maritime Writings of David M. Williams, edited by Lars Scholl, 81-108. St John’s, Newfoundland: International Maritime Economic History Association, 2000.

Wright, Charles and C. Ernest Fayle. A History of Lloyd’s. London: Macmillan 1928.

40 Table 1

Insurance Premium by Route, 1730-1830 Percent of Value Insured

New East Date Hamburg Riga Lisbon Leghorn Smyrna Canada Jamaica York Indies 1730-1731 1.05 . 1.05 1.50 1.50 . 2.00 2.50 3.20 1764-1770 1.06 1.74 1.16 1.51 1.64 3.36 2.53 1.51 3.50 1768-1770 1.00 . 1.05 1.50 1.75 . 2.10 2.50 3.68 1783-1785 1.10 1.20 1.10 1.50 2.10 3.10 2.10 2.50 4.20 1820-1829 0.40 0.50 0.80 2.00 1.20 1.80 1.20 1.50 3.00 1828 . 0.63 0.75 . . . 1.00 1.50 3.00 Sources. 1730-1731, 1768-1770: London Assurance Company (John 1951, Table 3). 1764-1770: William Braund’s Journal of Risks (Essex County Record Office, D/DRu B7, courtesy of Adrian Leonard). 1783-1785: Prince's Price Current. 1821-1829: London New Price Current. 1828: Wright and Fayle History of Lloyd's (1928, 319-320 )

41 Table 2

Ship losses in 1834 by source

Observations Average Share tonnage foreign

Registers 567 137 32.5 Lloyd's List 495 178 57.0 Wikipedia 560 171 51.8

Registers only 339 116 17.4 Lloyd's only 77 185 66.2 Wikipedia only 138 141 47.8 Registers and Lloyd's 30 179 66.7 Registers and Wikipedia 34 151 38.2 Lloyd's and Wikipedia 224 189 53.1 All three sources 164 170 56.1 Total distinct losses 1006 137 41.7

Sources: Lloyd's and Registers: United Kingdom (1836), pp. 307-314, 322, 348-377

42 Table 3

Losses and Loss Rates in Peacetime Periods

Losses Loss rates (number) (per cent) 1784-6 1824-6 1784-6 1824-6 Baltic 85 236 0.8 1.2 North Sea 38 117 0.2 0.4 Iberia & Biscay 53 68 0.6 0.3 Mediterranean 25 43 1.9 1.1 Canada 24 182 1.1 1.7 USA 56 46 2.4 1.1 West Indies & Africa 137 95 3.5 1.4 Asia 36 3.1

Notes: The denominator in each case is the larger of entries and clearances for each region. The regions are defined as follows: Baltic: Russia, Sweden, Prussia, East Country; North Sea: Denmark, Norway, Germany, Netherlands, Flanders; Biscay: France; Iberia: Spain, Portugal, Madeira, Canaries; Mediterranean: Gibraltar, Italy, Malta, Ionian Islands, Turkey and the Levant, Tripoli; Canada: British North America, Greenland, Davis Straits; United States: United States; West Indies: Coast of Africa, British West Indies, Hayti, Cuba, Mexico ; South America: Brazil, Columbia, Rio de la Plata, Chile, Peru, Whale Fisheries, Southern Fishery; Asia: Cape of Good Hope, St Helens, Mauritius, East Indies, China, Sumatra & Java, New South Wales, Van Diemen's Land, New Zealand. Sources: losses: Lloyd’s List, 1784-6, 1824-6; voyages: 1784-6: Statements of Navigation and Trade, National Archives, CUST 17/8-9; 1824-6: British Parliamentary Papers, Tables of the Revenue.

43 Figure 1

Number of arrivals in British and Irish ports by place of origin.

Sources: 1784-6: Statements of Navigation and Trade, National Archives, CUST 17/8-9; 1824-6: British Parliamentary Papers, Tables of the Revenue.

44 Figure 2. Insurance rates on European and oceanic routes (per cent of value insured)

Source: Table 1.

45 Figure 3. Percentage loss rates per voyage on European and oceanic routes

Sources: Table 3, except EIC (East India Company) for all voyages in 1783-91 and 1820-29 from Solar (2013).

46 Figure 4. Grenville Collins, Map of Dublin Bay, 1693.

Figure 5. Detail from Hamilton Moore’s “New and Correct Chart of the Baltic ...”, 1791.

47